VirtualBox 5.1.22 on Ubuntu 16.04, Ubuntu/Mate guest hangs and 100% CPU

[Shahim]

I’ve had this problem for a while (even on 4.*) and I finally was able to collect some needed information. The guest frequently hangs and I am not able to use the GUI, ssh, ping, etc. to connect. I have to poweroff the guest. I also notice that one of the cores/threads remains at 100% when this occurs. In htop I was also seeing an EMT-3 process that has the same CPU usage.

Some related problems I’ve noticed with this VM in case they are useful. I’ve noticed random desktop freezes while dragging windows around. I think they happen when the window snaps to borders or when they are moved from one screen to another. This freeze is just for the desktop and I am able to ssh to it and reboot. Another problem is when I use Windows 10 RDP to connect, sometimes I can’t reconnect after disconnect. If I remember correctly, this is a complete freeze that I can’t recover from without a poweroff.

This is for VirtualBox Version 5.1.22 r115126 (Qt5.5.1) originally installed as 4.* and then upgraded to 5.0 and 5.1, all from the VirtualBox debian xenial contrib repository. Oracle extension pack 5.1.22r115126.

The host is Ubuntu 16.04 64G RAM, ZFS filesystem, with latest updates. Uname: Linux main 4.4.0-77-generic #98-Ubuntu SMP Wed Apr 26 08:34:02 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

The guest is also Ubuntu 16.04 with Mate desktop. uname:Linux devvm 4.4.0-75-generic #96-Ubuntu SMP Thu Apr 20 09:56:33 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux The matching guest additions are installed. It is started with vboxautostart and saved/resumed on reboot. I think I haven’t seen the problem when I run the guest with 1 CPU but I’m not sure. I can almost say that the problem does occur with any other combination of VM settings. I’m currently trying 3 CPUs and no audio based on a comment in a forum thread discussing a similar problem. This problem also occurs if the guest is directly started from the VirtualBox GUI (instead of headless). I really need multiple CPUs for this VM to be useful for my needs.

Attached is a zip of the VM log, service log, vminfo, lspci, and cpuinfo.

I also have a core dump file but I’m not sure if it is a full dump. It is only 15M in size and the VM had 16G RAM when this happened. Does the core file have any sensitive information? Is there any other place to post it other than publicly to the forum?

Please let me know what additional information might be useful (and how to get it) and I’ll try to collect it the next time this happens. Thanks in advance for any help.

[Perryg]

Here's something you don't see everyday.

Code: Select all

00:00:10.295577 [/Devices/ahci/0/LUN#0/Config/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/] (level 45)

Why all the subfolders?

What happens if you select the piix3 instead of ich9 chipset in the motherboard section of the guest settings?

[socratis]

Why all the subfolders?

If I were to misquote an old TV series: "Snapshot, snapshot, snapshot!"

[Shahim]

Here's something you don't see everyday.

Code: Select all

00:00:10.295577 [/Devices/ahci/0/LUN#0/Config/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/Parent/] (level 45)

Why all the subfolders?

I do have about 40 nested snapshots for this vm on this run. Does this cause problems? I have a boot script on the host that takes snapshots and deletes old ones.

What happens if you select the piix3 instead of ich9 chipset in the motherboard section of the guest settings?

I'll switch the chipset the next time I reboot the guest. I can't really remember if I had the same problem while using that chipset.

[mpack]

I do have about 40 nested snapshots for this vm on this run. Does this cause problems?

It's interesting that you should use the words "snapshots" and "problems" in one post, as those are the words I often recommend are used in a Google search by people who wish to research the issue. I.e. Google search "snapshots problems site:forums.virtualbox.org".

[Perryg]

40 nested snapshots

Wow!

I believe I am not the one to continue here as I think snapshots are evil, and if used should be deleted after the need has been satisfied.

[socratis]

I do have about 40 nested snapshots for this vm on this run. Does this cause problems? I have a boot script on the host that takes snapshots and deletes old ones.

Does this cause problems? No, just as like lighting my cigarette next to the 1000 lt gasoline open container. It could go fine, until it doesn't. Then it's not an accident, it's a major disaster. Snapshots are not backups, they are serial links of a chain. So, if anything goes wrong to link 12 or link 27? Kaboom.

[Shahim]

Hi mpack,

I do have about 40 nested snapshots for this vm on this run. Does this cause problems?

It's interesting that you should use the words "snapshots" and "problems" in one post, as those are the words I often recommend are used in a Google search by people who wish to research the issue. I.e. Google search "snapshots problems site forums virtualbox org".

I don't want to misunderstand your reply. Are you saying that I shouldn't have asked such a simple question in reply to Perryg's statement that "this is not seen everyday" in a thread I started without doing a search? If this is the case, then I don't find your reply or attitude very friendly or helpful. This is the first time I post to this forum asking for help after I've spend 10s of hours researching old threads that sounded similar to my problem and trying the various (mostly semi-random) suggestions without any good results. Also, most of those threads also die without any clear resolution. Your comment would be valid if I post regularly to this form without doing any research.

My apologies if I misunderstood your intentions.

[Shahim]

Let me address the snapshots issue so that the main problem doesn't get lost in discussing snapshots (unless they are the cause of the hangs). The main reason I have so many snapshots at this point is because of the hangs I started to have with my guest in case I need to revert to an older state. I also know that they are not backups. I have a daily backups done on the VMs while they are poweredoff or saved (actually, a fast ZFS snapshot/send to NAS with every VM boot). I understand that if any one gets corrupted that whole disk is corrupted but unless they might explain the hangs, I do have a very reliable snapshot and backup routine that has worked so far. The snapshots are usually not more than 10 but the many VM reboots I did yesterday to catch a hang added to this high number.

I'm willing to remove all snapshots for now if I will not be able to get some help solving the freezing problems with them. I'm trying to be as helpful as I can to provide the needed information and taking the needed steps to identify the cause of this problem.

[Perryg]

I would wait and try what I asked first, but if it were me I would get rid off all snapsots as soon as I cold. But that's just me. Could they be the cause of the hangs? Maybe. I know that you can overrun paths and have all kinds of issues, but whether that is the cause here I have no clue.

[Shahim]

Perryg, thanks for your suggestions. I removed all snapshots without problems, changed to PIIX3, 4 CPU, and ALSA/ICH audio. Took one snapshot after these changes.

Then after starting the guest with vboxautostart and connecting to it with the VirtualBox GUI, the VM went into guru meditation. Attached are the logs and debug information. The VM is still in GM. Is there any other information I can get before using kill -4 (to get a core dump as described in the docs)?

[Perryg]

"VCPU3: Guru Meditation -2701 (VERR_VMM_RING0_ASSERTION)" is a serious error and one that AFAIK does not have an easy answer as far as knowing what the cause and fix is.

Question: does the assertion stop if you change back to ich9? or is this error something you see sometimes regardless?

[Shahim]

The VM is still in GM. I'll try changing to ich9 after I reboot it.

I'm not that familiar with troubleshooting VirtualBox but I've just noticed some USB hardware issues on the host that I should probably figure out before worrying about the VirtualBox problems. There are many log messages in the host's syslog about the kernel having intermittent problems with one or more of the USB devices.

[Perryg]

Not a bad idea. I was leaning to a host issue whether being hardware or host side code. I do remember a time or two when these kind of things were happening where a bios update was needed and on another occasion it was due to the hardware being HP with a specific CPU and had some issue I don't remember off the top of my head, but perhaps you can investigate this further and see if it has a baring.

[mpack]

I don't want to misunderstand your reply.

And yet you did, and ran with it. My post meant no more than was explicit in the words, which was: this search shows you what problems heavy use of snapshots can cause.